The present invention relates to a network system and its software management method.
As computer application expands, the range affected by faulty operation of computer becomes wider and wider. Therefore, system and software production with high reliability is intensely demanded as a very important subject. For example, in computer systems for control, controlled systems are typically machines and facilities and a slight delay or a slight control mistake might cause a great deal of damage. In some cases, accidents may result in injury or death. High reliability and high responsiveness are thus demanded. Therefore, highly reliable software production (software development and maintenance) is intensely demanded.
On the other hand, highly reliable software production (software development and maintenance) needs sufficient check and time, efficiency improvement in development and maintenance being hindered. As a result of pursuit of high reliability excessively, software development and maintenance often lack expandability and flexibility. As a result, highly reliable software increases the system cost. At the time of system enlargement or reconstruction, a further enormous investment is needed for software development and maintenance, resulting in problems of insufficient efficiency improvement, expandability, and flexibility.
In this regard, JP-B-2-23887 discloses a technique whereby system stop caused by software maintenance is avoided in one computer by making possible maintenance of program in operation.
However, application to the entire network system is not considered.
In the above described prior art, application to the entire network system is not considered. For the purpose of pursuit of higher reliability, therefore, nonstop replacement is applied to the entire network system uniformly, resulting in an increased operational cost and increased replacement program storage areas. Further, since application to the entire network system is not considered, the program maintenance method of the entire network system is indefinite. This results in problems such as occurrence of program loading mistakes with a computer taken as the unit. The cost of production of a highly reliable network system in development and maintenance is increased as higher reliability is pursued and as the scale of the network system is expanded. Further, the network system lacks flexibility and expandability in some cases.